Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 1 | //===- Wasm.h - Wasm object file format -------------------------*- C++ -*-===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This file defines manifest constants for the wasm object file format. |
| 11 | // See: https://github.com/WebAssembly/design/blob/master/BinaryEncoding.md |
| 12 | // |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | #ifndef LLVM_BINARYFORMAT_WASM_H |
| 16 | #define LLVM_BINARYFORMAT_WASM_H |
| 17 | |
| 18 | #include "llvm/ADT/ArrayRef.h" |
| 19 | |
| 20 | namespace llvm { |
| 21 | namespace wasm { |
| 22 | |
| 23 | // Object file magic string. |
| 24 | const char WasmMagic[] = {'\0', 'a', 's', 'm'}; |
| 25 | // Wasm binary format version |
| 26 | const uint32_t WasmVersion = 0x1; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 27 | // Wasm linking metadata version |
| 28 | const uint32_t WasmMetadataVersion = 0x1; |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 29 | // Wasm uses a 64k page size |
| 30 | const uint32_t WasmPageSize = 65536; |
| 31 | |
| 32 | struct WasmObjectHeader { |
| 33 | StringRef Magic; |
| 34 | uint32_t Version; |
| 35 | }; |
| 36 | |
| 37 | struct WasmSignature { |
| 38 | std::vector<uint8_t> ParamTypes; |
| 39 | uint8_t ReturnType; |
| 40 | }; |
| 41 | |
| 42 | struct WasmExport { |
| 43 | StringRef Name; |
| 44 | uint8_t Kind; |
| 45 | uint32_t Index; |
| 46 | }; |
| 47 | |
| 48 | struct WasmLimits { |
| 49 | uint8_t Flags; |
| 50 | uint32_t Initial; |
| 51 | uint32_t Maximum; |
| 52 | }; |
| 53 | |
| 54 | struct WasmTable { |
| 55 | uint8_t ElemType; |
| 56 | WasmLimits Limits; |
| 57 | }; |
| 58 | |
| 59 | struct WasmInitExpr { |
| 60 | uint8_t Opcode; |
| 61 | union { |
| 62 | int32_t Int32; |
| 63 | int64_t Int64; |
| 64 | int32_t Float32; |
| 65 | int64_t Float64; |
| 66 | uint32_t Global; |
| 67 | } Value; |
| 68 | }; |
| 69 | |
| 70 | struct WasmGlobalType { |
| 71 | uint8_t Type; |
| 72 | bool Mutable; |
| 73 | }; |
| 74 | |
| 75 | struct WasmGlobal { |
| 76 | uint32_t Index; |
| 77 | WasmGlobalType Type; |
| 78 | WasmInitExpr InitExpr; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 79 | StringRef SymbolName; // from the "linking" section |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 80 | }; |
| 81 | |
| 82 | struct WasmImport { |
| 83 | StringRef Module; |
| 84 | StringRef Field; |
| 85 | uint8_t Kind; |
| 86 | union { |
| 87 | uint32_t SigIndex; |
| 88 | WasmGlobalType Global; |
| 89 | WasmTable Table; |
| 90 | WasmLimits Memory; |
| 91 | }; |
| 92 | }; |
| 93 | |
| 94 | struct WasmLocalDecl { |
| 95 | uint8_t Type; |
| 96 | uint32_t Count; |
| 97 | }; |
| 98 | |
| 99 | struct WasmFunction { |
| 100 | uint32_t Index; |
| 101 | std::vector<WasmLocalDecl> Locals; |
| 102 | ArrayRef<uint8_t> Body; |
| 103 | uint32_t CodeSectionOffset; |
| 104 | uint32_t Size; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 105 | uint32_t CodeOffset; // start of Locals and Body |
| 106 | StringRef SymbolName; // from the "linking" section |
| 107 | StringRef DebugName; // from the "name" section |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 108 | uint32_t Comdat; // from the "comdat info" section |
| 109 | }; |
| 110 | |
| 111 | struct WasmDataSegment { |
| 112 | uint32_t MemoryIndex; |
| 113 | WasmInitExpr Offset; |
| 114 | ArrayRef<uint8_t> Content; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 115 | StringRef Name; // from the "segment info" section |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 116 | uint32_t Alignment; |
| 117 | uint32_t Flags; |
| 118 | uint32_t Comdat; // from the "comdat info" section |
| 119 | }; |
| 120 | |
| 121 | struct WasmElemSegment { |
| 122 | uint32_t TableIndex; |
| 123 | WasmInitExpr Offset; |
| 124 | std::vector<uint32_t> Functions; |
| 125 | }; |
| 126 | |
| 127 | // Represents the location of a Wasm data symbol within a WasmDataSegment, as |
| 128 | // the index of the segment, and the offset and size within the segment. |
| 129 | struct WasmDataReference { |
| 130 | uint32_t Segment; |
| 131 | uint32_t Offset; |
| 132 | uint32_t Size; |
| 133 | }; |
| 134 | |
| 135 | struct WasmRelocation { |
| 136 | uint8_t Type; // The type of the relocation. |
| 137 | uint32_t Index; // Index into either symbol or type index space. |
| 138 | uint64_t Offset; // Offset from the start of the section. |
| 139 | int64_t Addend; // A value to add to the symbol. |
| 140 | }; |
| 141 | |
| 142 | struct WasmInitFunc { |
| 143 | uint32_t Priority; |
| 144 | uint32_t Symbol; |
| 145 | }; |
| 146 | |
| 147 | struct WasmSymbolInfo { |
| 148 | StringRef Name; |
| 149 | uint8_t Kind; |
| 150 | uint32_t Flags; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 151 | StringRef Module; // For undefined symbols the module name of the import |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 152 | union { |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 153 | // For function or global symbols, the index in function or global index |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 154 | // space. |
| 155 | uint32_t ElementIndex; |
| 156 | // For a data symbols, the address of the data relative to segment. |
| 157 | WasmDataReference DataRef; |
| 158 | }; |
| 159 | }; |
| 160 | |
| 161 | struct WasmFunctionName { |
| 162 | uint32_t Index; |
| 163 | StringRef Name; |
| 164 | }; |
| 165 | |
| 166 | struct WasmLinkingData { |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 167 | uint32_t Version; |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 168 | std::vector<WasmInitFunc> InitFunctions; |
| 169 | std::vector<StringRef> Comdats; |
| 170 | std::vector<WasmSymbolInfo> SymbolTable; |
| 171 | }; |
| 172 | |
| 173 | enum : unsigned { |
| 174 | WASM_SEC_CUSTOM = 0, // Custom / User-defined section |
| 175 | WASM_SEC_TYPE = 1, // Function signature declarations |
| 176 | WASM_SEC_IMPORT = 2, // Import declarations |
| 177 | WASM_SEC_FUNCTION = 3, // Function declarations |
| 178 | WASM_SEC_TABLE = 4, // Indirect function table and other tables |
| 179 | WASM_SEC_MEMORY = 5, // Memory attributes |
| 180 | WASM_SEC_GLOBAL = 6, // Global declarations |
| 181 | WASM_SEC_EXPORT = 7, // Exports |
| 182 | WASM_SEC_START = 8, // Start function declaration |
| 183 | WASM_SEC_ELEM = 9, // Elements section |
| 184 | WASM_SEC_CODE = 10, // Function bodies (code) |
| 185 | WASM_SEC_DATA = 11 // Data segments |
| 186 | }; |
| 187 | |
| 188 | // Type immediate encodings used in various contexts. |
| 189 | enum : unsigned { |
| 190 | WASM_TYPE_I32 = 0x7F, |
| 191 | WASM_TYPE_I64 = 0x7E, |
| 192 | WASM_TYPE_F32 = 0x7D, |
| 193 | WASM_TYPE_F64 = 0x7C, |
| 194 | WASM_TYPE_ANYFUNC = 0x70, |
| 195 | WASM_TYPE_EXCEPT_REF = 0x68, |
| 196 | WASM_TYPE_FUNC = 0x60, |
| 197 | WASM_TYPE_NORESULT = 0x40, // for blocks with no result values |
| 198 | }; |
| 199 | |
| 200 | // Kinds of externals (for imports and exports). |
| 201 | enum : unsigned { |
| 202 | WASM_EXTERNAL_FUNCTION = 0x0, |
| 203 | WASM_EXTERNAL_TABLE = 0x1, |
| 204 | WASM_EXTERNAL_MEMORY = 0x2, |
| 205 | WASM_EXTERNAL_GLOBAL = 0x3, |
| 206 | }; |
| 207 | |
| 208 | // Opcodes used in initializer expressions. |
| 209 | enum : unsigned { |
| 210 | WASM_OPCODE_END = 0x0b, |
| 211 | WASM_OPCODE_GET_GLOBAL = 0x23, |
| 212 | WASM_OPCODE_I32_CONST = 0x41, |
| 213 | WASM_OPCODE_I64_CONST = 0x42, |
| 214 | WASM_OPCODE_F32_CONST = 0x43, |
| 215 | WASM_OPCODE_F64_CONST = 0x44, |
| 216 | }; |
| 217 | |
| 218 | enum : unsigned { |
| 219 | WASM_LIMITS_FLAG_HAS_MAX = 0x1, |
| 220 | }; |
| 221 | |
| 222 | // Subset of types that a value can have |
| 223 | enum class ValType { |
| 224 | I32 = WASM_TYPE_I32, |
| 225 | I64 = WASM_TYPE_I64, |
| 226 | F32 = WASM_TYPE_F32, |
| 227 | F64 = WASM_TYPE_F64, |
| 228 | EXCEPT_REF = WASM_TYPE_EXCEPT_REF, |
| 229 | }; |
| 230 | |
| 231 | // Kind codes used in the custom "name" section |
| 232 | enum : unsigned { |
| 233 | WASM_NAMES_FUNCTION = 0x1, |
| 234 | WASM_NAMES_LOCAL = 0x2, |
| 235 | }; |
| 236 | |
| 237 | // Kind codes used in the custom "linking" section |
| 238 | enum : unsigned { |
| 239 | WASM_SEGMENT_INFO = 0x5, |
| 240 | WASM_INIT_FUNCS = 0x6, |
| 241 | WASM_COMDAT_INFO = 0x7, |
| 242 | WASM_SYMBOL_TABLE = 0x8, |
| 243 | }; |
| 244 | |
| 245 | // Kind codes used in the custom "linking" section in the WASM_COMDAT_INFO |
| 246 | enum : unsigned { |
| 247 | WASM_COMDAT_DATA = 0x0, |
| 248 | WASM_COMDAT_FUNCTION = 0x1, |
| 249 | }; |
| 250 | |
| 251 | // Kind codes used in the custom "linking" section in the WASM_SYMBOL_TABLE |
| 252 | enum WasmSymbolType : unsigned { |
| 253 | WASM_SYMBOL_TYPE_FUNCTION = 0x0, |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 254 | WASM_SYMBOL_TYPE_DATA = 0x1, |
| 255 | WASM_SYMBOL_TYPE_GLOBAL = 0x2, |
| 256 | WASM_SYMBOL_TYPE_SECTION = 0x3, |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 257 | }; |
| 258 | |
| 259 | const unsigned WASM_SYMBOL_BINDING_MASK = 0x3; |
| 260 | const unsigned WASM_SYMBOL_VISIBILITY_MASK = 0xc; |
| 261 | |
| 262 | const unsigned WASM_SYMBOL_BINDING_GLOBAL = 0x0; |
| 263 | const unsigned WASM_SYMBOL_BINDING_WEAK = 0x1; |
| 264 | const unsigned WASM_SYMBOL_BINDING_LOCAL = 0x2; |
| 265 | const unsigned WASM_SYMBOL_VISIBILITY_DEFAULT = 0x0; |
| 266 | const unsigned WASM_SYMBOL_VISIBILITY_HIDDEN = 0x4; |
| 267 | const unsigned WASM_SYMBOL_UNDEFINED = 0x10; |
| 268 | |
| 269 | #define WASM_RELOC(name, value) name = value, |
| 270 | |
| 271 | enum : unsigned { |
| 272 | #include "WasmRelocs.def" |
| 273 | }; |
| 274 | |
| 275 | #undef WASM_RELOC |
| 276 | |
| 277 | // Useful comparison operators |
| 278 | inline bool operator==(const WasmSignature &LHS, const WasmSignature &RHS) { |
| 279 | return LHS.ReturnType == RHS.ReturnType && LHS.ParamTypes == RHS.ParamTypes; |
| 280 | } |
| 281 | |
| 282 | inline bool operator!=(const WasmSignature &LHS, const WasmSignature &RHS) { |
| 283 | return !(LHS == RHS); |
| 284 | } |
| 285 | |
| 286 | inline bool operator==(const WasmGlobalType &LHS, const WasmGlobalType &RHS) { |
| 287 | return LHS.Type == RHS.Type && LHS.Mutable == RHS.Mutable; |
| 288 | } |
| 289 | |
| 290 | inline bool operator!=(const WasmGlobalType &LHS, const WasmGlobalType &RHS) { |
| 291 | return !(LHS == RHS); |
| 292 | } |
| 293 | |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame^] | 294 | std::string toString(wasm::WasmSymbolType type); |
| 295 | std::string relocTypetoString(uint32_t type); |
| 296 | |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 297 | } // end namespace wasm |
| 298 | } // end namespace llvm |
| 299 | |
| 300 | #endif |